JENSEN, RICHARD; PHD

                         NORTHWESTERN UNIVERSITY, 1980
 
                         ECONOMICS, THEORY (0511)
 

                         Optimal behavior of a firm toward an exogenously developed innovation of uncertain profitability is
                         examined. It is shown that a firm may delay adopting a profitable innovation if it is uncertain about the
                         innovation's profitability but can reduce the uncertainty by waiting and gathering information. The
                         decision problem is formalized as an optimal stopping problem in which at each decision date the firm can
                         either adopt or wait and take an observation. The firm starts with a subjective probabilistic belief that the
                         innovation will be profitable which is updated according to generalized learning rules after each
                         observation. The optimal decision rule can be characterized by a unique reservation probability
                         representing the minimum value which the firm's current (updated) belief that the innovation will be
                         profitable must attain to induce adoption. When a delay in adoption occurs, that delay will be longer when
                         the firm's original belief is smaller, the information received is less favorable, the ability to learn is
                         unlimited, and the expected profitability is lower. An industry model in the absence of rivalry is
                         constructed in which firms are Bayesian learners and are identical in all respects except for their original
                         beliefs. When these beliefs are distributed uniformly over the unit interval, the expected diffusion curve
                         will be either S-shaped or concave. Rivalry is examined in a duopoly model in which the existence of a
                         Bayesian (fulfilled expectations) equilibrium is proved. In it each firm accurately and endogenously
                         estimates the probability of adoption by the rival. Adoption by one firm need not increase the probability
                         of adoption by the other firm. These last two results raise serious questions about the assumptions
                         commonly made in empirical studies of diffusion.

Social Systems Simulation Group P.O. Box 6904 San Diego, CA  92166-0904 Roland Werner, Principal Phone/FAX  (619) 660-1603